| Abstract | In nine lexical decision experiments we evaluated Berent and Perfetti’s (1995) hypothesis that, in visual word recognition, consonants and vowels are derived in two consecutive cycles. A nonviable prime of 14 ms duration within a sequence of mask, nonword prime, and word target provided the operation definition of the first cycle. Contrary to the hypothesis, within the first cycle psuedohomophones (e.g., KLIP-clip) primed better than nonwords that conveyed only consonant information (e.g., CLEP-clip), vowel-preserving primes (e.g. GLEZE-glaze, BAK-bad) were as effective as consonant-preserving primes (e.g., GLEZE-glaze, BYD-bad) and vowel complexity seemed to matter. Discussion focused on the assembly of phonology and the immediacy and primacy of phonological codes. |